Long-term Follow-up of a Randomized Clinical Trial Assessing the Efficacy of Medication, Acupuncture, and Spinal Manipulation for Chronic Mechanical Spinal Pain Syndromes


Long-term Follow-up of a Randomized Clinical Trial Assessing the Efficacy of Medication, Acupuncture, and Spinal Manipulation for Chronic Mechanical Spinal Pain Syndromes

Reinhold Muller, PhD, Lynton G.F. Giles, DC, PhD

Lynton G.F. Giles, DC, PhD,
PO Box 880, Indooroopilly,
Queensland 4068, Australia

  To assess the long-term benefits of medication, needle acupuncture, and spinal manipulation as exclusive and standardized treatment regimens in patients with chronic (>13 weeks) spinal pain syndromes.

STUDY DESIGN:   Extended follow-up (>1 year) of a randomized clinical trial was conducted at the multidisciplinary spinal pain unit of Townsville’s General Hospital between February 1999 and October 2001.

PATIENTS AND METHODS:   Of the 115 patients originally randomized, 69 had exclusively been treated with the randomly allocated treatment during the 9-week treatment period (results at 9 weeks were reported earlier). These patients were followed up and assessed again 1 year after inception into the study reapplying the same instruments (ie, Oswestry Back Pain Index, Neck Disability Index, Short-Form-36, and Visual Analogue Scales). Questionnaires were obtained from 62 patients reflecting a retention proportion of 90%. The main analysis was restricted to 40 patients who had received exclusively the randomly allocated treatment for the whole observation period since randomization.

RESULTS:   Comparisons of initial and extended follow-up questionnaires to assess absolute efficacy showed that only the application of spinal manipulation revealed broad-based long-term benefit: 5 of the 7 main outcome measures showed significant improvements compared with only 1 item in each of the acupuncture and the medication groups.

CONCLUSIONS:   In patients with chronic spinal pain syndromes, spinal manipulation, if not contraindicated, may be the only treatment modality of the assessed regimens that provides broad and significant long-term benefit.

The results of this “fastidious” approach were able to add some information regarding the efficacy of treatment regimens in patients with chronic spinal pain syndromes. Overall, patients who have chronic mechanical spinal pain syndromes and received spinal manipulation gained significant broad-based beneficial short-term and long-term outcomes. For patients receiving acupuncture, consistent improvements were also observed, although without reaching statistical significance (with a single exception). For patients receiving medication, the findings were less favorable. Larger studies are now clearly justified.

From the Full-Text Article:


This is, to the authors’ knowledge, the first report on long-term efficacy of 3 distinct and standardized treatment regimens for patients with chronic spinal pain syndromes using a “fastidious” approach; that is, the only type of study from which potentially valid inferences of cause and effect can directly be drawn. [24] The validity of the study (ie, the absence of different types of bias) is hereby essential and will be discussed first.

Selection Bias

The study sample has a broad socioeconomic background and a wide age range. Quite stringent exclusion criteria guaranteed a pathologically homogeneous sample. It was successfully ascertained that all “dropouts” occurring during the study treatment period, as well as during the extended follow-up period thereafter, occurred for reasons unrelated to the study outcome (ie, moving overseas, being transferred, etc). A high retention proportion of 90% for this extended follow-up study, together with the above stated facts, supports the generalizability of the findings.

Information Bias

Intention-to-treat analyses including noncompliers (1 for the 9-week treatment period [9] and 1 for the presented study) revealed results quite consistent with the respective compliers-only analyses thus effectively diminishing any relevant misclassification bias from noncompliers. A different color code was used from that in the 9-week analysis to ensure successful blinding [25] of data analysis. All data handling and analyses were again performed before the treatment color code was broken. The senior biostatistician was involved neither in the data collection process nor in any daily business of the center, thus minimizing information bias.

Blinding of the physician was not possible; even if, for instance, a “sham” acupuncture treatment would have been regarded as ethically justifiable, the acupuncturist would still have to know what treatment to perform. Blinding of the patients was not possible because there is, for instance, no known practical way to perform a sham manipulation. The potential for information bias, in this context, however, seems limited by the standardized treatment regimens and the fact that the clinician was not involved in measuring outcome. Information bias arising from a placebo effect or from a self-limiting effect is highly unlikely because patients in this study had chronic spinal pain syndromes (the average duration of having this exceeded 2 years) and had long histories of having sought pain relief. Improvement caused by the abovementioned effects could be expected in cases with acute spinal pain [26] but seems rather unlikely in long-term cases.

Confounding Bias

Table 1 indicates that the 3 groups were very similar in their characteristics at inception. Additional bivariate and multivariate analyses of potential effects of these characteristics on the outcome measures also disproved any relevant confounding bias.

According to Turk and Rudy, [27] no clinical study can be completely valid because of the complexities of extended follow-up trials; however, we have attempted to conduct a well-executed extended follow-up randomized trial with a rigorous protocol, and the overall validity of the reported findings does not seem to be negatively affected by any obvious bias. The main emphasis of this study was to assess absolute efficacy; consequently, within-group comparisons constituted the basis for analysis. Additional across-group comparisons, as often used in clinical trials to assess relative efficacy (eg, when new treatments are compared with an accepted “gold” standard), would have resulted in 2-dimensional testing (ie, determining and validating a gold standard within the same data set) defying any meaningful interpretation. The validation process (ie, the relative comparison) consequently has to be reserved for future trials.

However, the presented trial with successful randomization, thorough concealment, and within-group analyses applied the most powerful design possible to a research area where no accepted gold standard exists and where the emphasis, at this early stage of the research process, has to be on absolute, as opposed to relative, efficacy.

It should be noted that definitions of chronicity for low-back pain have been suggested by various authors such as Nachemson and Bigos [28] and by Skouen et al. [29] For the reported study, the definition for chronic pain duration was more than 13 weeks, so it is against this definition that these results are to be interpreted.

The overall results of this extended follow-up efficacy study appear to favor the application of manipulation and suggest that manipulation, if not contraindicated, and, to some extent, also needle acupuncture seem to successfully achieve long-term benefits in chronic spinal pain syndrome patients. However, no such benefit could be observed for medication. These results not only corroborate the findings of the 9-week analysis [9] but also of the smaller pilot study. [12] It seems noteworthy that the comparison of the percentages of those who had to change the treatment modality (because of side effects or unsatisfactory results) also appears to favor manipulation in that manipulation showed by far the lowest proportion (38.7%) of changeovers compared with acupuncture (53.3%) and medication (81.2%). Consequently, spinal manipulation appeared to provide the highest satisfaction. Moreover, both the 9-week findings and the extended follow-up results are consistent with conclusions by Meade et al, [10], [13] who, on comparing chiropractic with hospital therapists for treating low-back pain as they would in day-to-day practice (“pragmatic” approach), reported that those treated by chiropractic derived more short-term and long-term benefit and satisfaction than those treated by hospital therapists.

Medication apparently did not achieve an improvement in chronic spinal pain, although the SF-36 indicator of general health status did show an improvement (P = .02) for general health status. This may reflect some satisfaction with not having the inconvenience of needing to attend twice weekly for treatment and/or may also suggest that medication did not act as a nocebo.

It is interesting that the application of manipulation and acupuncture seem roughly equally successful in the ITT analysis, but only manipulation seems of broad-based long-term benefit in the compliers-only analysis. A more detailed look at the noncompliers data revealed that 4 of the 6 patients in the acupuncture arm who had some other type of treatment than the randomly allocated regimen during the extended follow-up period were actually treated with manipulation. Therefore, an artificial inflation of the effect of acupuncture treatment in the ITT analysis by additional manipulation therapy seems likely. The compliers-only analysis therefore seems to provide information that is more accurate.

The ITT analysis, however, is per se relevant because it displays the information that would be available from a similar trial in a larger metropolitan setting where the information on additional treatment may not be collected (or at least only less reliably). The setting of the present trial in a small, geographically relatively isolated community which is served by only 1 major public (providing free treatment) hospital rendered it possible to directly collect precise information on possible additional treatments during the extended follow-up period by checking the single public hospital’s computer records.

This advantage of the small community setting, however, is partly offset by a long inception period (several years) to reach the minimum necessary sample size. In this context, it seems noteworthy that because of the necessarily stringent inclusion and exclusion criteria, 533 patients had to be seen (and treated) at the unit to achieve the reported sample sizes, reflecting that only around 1 (22.3%) of 5 patients fulfilled the inclusion /exclusion criteria.

Another general reason for the relatively small sample sizes for the extended follow-up analysis, however, lies in the very nature of this strictly fastidious approach itself: the group of strict compliers necessarily dwindles with increasing period of observation as the likelihood increases that additional treatment (eg, simple pain killers) is used by those in the long-term condition. This consequence of the fastidious approach, however, is easily compensated for by the fact that it is the only approach where an observed effect can be unambiguously attributed to 1 specific treatment modality only (if the study follows an otherwise rigorous methodology). Moreover, it seems worth reiterating that statistical testing takes into account the sample size and the observed effects proved to be both medically relevant and statistically significant.

It should be emphasized that this study was exclusively concerned with chronic spinal pain, and therefore, no statement whatsoever can be made about the potential role of the investigated regimens in treating acute spinal pain syndromes.


Chronic mechanical spinal pain syndromes are prevalent conditions [30] that tend to create a cluster of related problems reaching from withdrawal from social activity to a compromised immune function. [31] The associated resulting direct and indirect costs in industrialized communities are vast. [32] A large community study seems to be the next logical step to address this important problem and to further investigate the reported findings. Consideration should also be given to assessing the efficacy of other treatment modalities. This suggested study should be based on a fastidious approach and incorporate an expanded multidisciplinary team to gain further evidence-based information on the absolute and also the relative efficacy of all forms of available treatments.

The results of this “fastidious” approach were able to add some information regarding the efficacy of treatment regimens in patients with chronic spinal pain syndromes. Overall, patients who have chronic mechanical spinal pain syndromes and received spinal manipulation gained significant broad-based beneficial short-term and long-term outcomes. For patients receiving acupuncture, consistent improvements were also observed, although without reaching statistical significance (with a single exception). For patients receiving medication, the findings were less favorable. Larger studies are now clearly justified.